INDEX

Explanations

references to tape and related products

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

躇

-0.46

 Util

-0.39

rž

-0.38

Ʋ

-0.38

 Osorio

-0.38

không

-0.35

ồm

-0.35

 Prosper

-0.35

ris

-0.34

 moradores

-0.34

POSITIVE LOGITS

 Tape

1.31

 tape

1.30

 TAPE

1.25

Tape

1.24

tape

1.07

 tapes

1.04

 taped

0.89

 Tapes

0.86

tapes

0.81

 テープ

0.81

Activations Density 0.004%

references to tape and related products

No Comments

No Known Activations